Does the Summarize Tool read footnotes?
Does the Summarize tool in the resource panel read the footnotes in the selected text and take that information into account in its summarization? Or does it ignore footnotes and just summarize the main text? (See below for what I'm referring to.)
Is this summary taking into account the information in footnotes 29 and 30 in the final paragraph?
Comments
-
No. Footnotes are not included in the summarized content.
0 -
Alan Palmer (Logos) said:
No. Footnotes are not included in the summarized content.
Follow-up question: What is the maximum word count for these summaries? For example, I just summarized an entire part of a book, Part III of this book contained two chapters. I copied the summary into MS Word and it was 269 words long. What would be the maximum allowable word count for the summaries? Is it something like 300 words?
0 -
Kiyah said:
Follow-up question: What is the maximum word count for these summaries?
The maximum length is based on the large language model backing the feature. As of today, summaries are generated by a model that has a max context of 128,000 tokens.
0 -
Alan Palmer (Logos) said:
The maximum length is based on the large language model backing the feature. As of today, summaries are generated by a model that has a max context of 128,000 tokens.
Does the token count for the summarisation include the word count of the text to be summarised as well as the word count of the summary? I think the consumption of tokens included the work done to produce the summary, as well as the production of the summary itself. Is that right?
👁️ 👁️
0 -
Alan Palmer (Logos) said:
As of today, summaries are generated by a model that has a max context of 128,000 tokens.
Where can I find out what this means? In the article on How do AI Credits work, nothing is mentioned about tokens. Why is it so difficult to understand how much of AI we are using each time we us it?
0 -
Mark said:
Why is it so difficult to understand how much of AI we are using each time we us it?
We intentionally designed the limits to be generous and hope users never have to think about it. If you are exclusively using Summarization, current limits give you thousands of uses each month! We want you to use them freely and not be bothered by the credit usage or not.
At this time, limits are only there as a type of fair use cap to avoid abuse. We monitor usage across our applications and reserve the right to adjust the limits as we need to in order to best support our community of users.
Admittedly, Smart Synopsis is quite "expensive" relative to the other features. We are actively thinking about ways to fix that without compromising the quality and value of that feature. We will be evaluating how it is used in the first months of the full subscription launch.
0 -
Antony Brennan said:
Does the token count for the summarisation include the word count of the text to be summarised as well as the word count of the summary?
That's right. However, the prompt overhead is often small relative to the content being summarized.
0 -
Mark said:Alan Palmer (Logos) said:
As of today, summaries are generated by a model that has a max context of 128,000 tokens.
Where can I find out what this means? In the article on How do AI Credits work, nothing is mentioned about tokens. Why is it so difficult to understand how much of AI we are using each time we us it?
I think it's because it isn't straightforward. The number of tokens used depends on various factors which are not always constant, so measuring them is difficult.
It's like measuring the consumption of petrol when driving a car. The manual for the car says that it is X miles per gallon, but in reality, it depends on how fast you're going, whether you're going uphill or downhill, according to some the weather is a factor. Then there is also the efficiency of the engine and the kind of fuel you're using. We all accept the average given in the manual as a fact, but if measured day to day it would change continually.
With the AI summary it's dependent on the length of the text being summarised, as well as the complexity of the language. It's also dependent on the length of the summary, and the effort required to produce coherent sentences.
Amongst the major AI providers, I haven't found anyone who will commit to how many tokens it takes to produce an outcome.
👁️ 👁️
0 -
Alan Palmer (Logos) said:
That's right. However, the prompt overhead is often small relative to the content being summarized.
Thanks
👁️ 👁️
0 -
Logos abstracts all of that detail away. Our users don't have to worry about any of it. You're free to just summarize your books to your heart's content.
We are essentially managing all those factors and differences for you. It all averages out across our user base and ideally nobody should need to worry about any limits (in reality, far less than 1% of our users ever even approach their monthly limits).
I understand things can change as we bring on more subscribers and continue to ship new and exciting features; but for now, I think the current limits should be more than adequate and stay out of your way. We will be sure to communicate if any changes need to be made and welcome your feedback along the way.
0 -
Alan Palmer (Logos) said:
Logos abstracts all of that detail away. Our users don't have to worry about any of it. You're free to just summarize your books to your heart's content.
I am in the 1%. It was the Synopsis. I felt compelled to see what it could do. It can write poems and perform other tricks.
What about generating questions in the Sermon Builder and in the Study Bible tool. Are they on the expensive side like Synopsis. Just looking for ways to manage overall consumption. It is good to hear I don't have to worry much about summarisation.
👁️ 👁️
0 -
Alan Palmer (Logos) said:
We intentionally designed the limits to be generous and hope users never have to think about it.
So, why the difference between Logos Pro and Max? The concern I have is that some have claimed to have come near the maximum amount of credit allowed per month. I don't know enough on how that was done if the limits are generous. If I use logos everyday and am doing searches constantly and using the summarization tool a lot or the translation or the smart searches, at the moment we are blind in knowing how many credits we are using. Do I need Max or Pro? Should l limit my use of AI when I really don't need to use it? It seems AI is being used all the time, especially when I do an all search (which by the way, I have not enjoyed that I cannot turn it off on an all search).
0 -
Mark said:
which by the way, I have not enjoyed that I cannot turn it off on an all search).
Actually, adding an AND or OR has the effect of turning it off
Orthodox Bishop Alfeyev: "To be a theologian means to have experience of a personal encounter with God through prayer and worship."; Orthodox proverb: "We know where the Church is, we do not know where it is not."
0 -
Antony Brennan said:
What about generating questions in the Sermon Builder and in the Study Bible tool. Are they on the expensive side like Synopsis. Just looking for ways to manage overall consumption. It is good to hear I don't have to worry much about summarisation.
Generating questions is even less of an issue than summarization.
Synopsis is uniquely expensive. As I mentioned earlier, we are considering many options to help users not hit their limit from this feature. We may change the default to only execute it on demand (it's currently defaulted on and stays on until users close it). We are constantly evaluating new AI models and adjusting our limits as third-party costs change.
0 -
Mark said:
It seems AI is being used all the time, especially when I do an all search (which by the way, I have not enjoyed that I cannot turn it off on an all search).
You can at least close the synopsis box and it remains closed until you decide to manually run a synopsis of your search results.
And as MJ mentions, you automatically switch from Smart search to Precise search in the All tab when you enter traditional Logos search operators. Not ideal, I wish you can turn it to Precise when you just want it to search for terms in resources without search operators. I also would want to change it to Precise get continuous scroll in the All tab, which is disabled when you do a Smart search, presumably to save credits.
0 -
Mark said:
So, why the difference between Logos Pro and Max?
The difference is the number of features that use the credits. Max adds translation. We include more credits so that users with more features can use them all as much as users on lower tiers. (simply put: more features = more credits)
Mark said:The concern I have is that some have claimed to have come near the maximum amount of credit allowed per month. I don't know enough on how that was done if the limits are generous.
So far, the most common reason has been extensive use of resource translation. It is an extremely popular feature among a very small cohort of users. This has been true since launching the feature in L10.
Mark said:If I use logos everyday and am doing searches constantly and using the summarization tool a lot or the translation or the smart searches, at the moment we are blind in knowing how many credits we are using.
The app will currently notify you when you reach 80% usage. We are considering changing this to show it earlier to give users like yourself greater peace of mind.
Mark said:Do I need Max or Pro?
I would advise you to choose based on the features each level offers rather than the credits. If you find you are running out, I would love to know since this is definitely something we want to minimize.
Mark said:Should l limit my use of AI when I really don't need to use it?
No. In fact, mindfulness of your limits limiting your usage of these features is something we want to avoid.
I will make one suggestion here that partially contradict that advice. If you are running a lot of searches, go ahead and turn off that synopsis card if you're not finding it useful.
Along these lines, we have made recent changes to our mobile apps to try to minimize inadvertent uses of the translation feature because we discovered a very small cohort of users were showing translations in the selection menu when making selections for other purposes.
These are both potential inadvertent uses. As we find these things, we will work to remedy them.
Mark said:It seems AI is being used all the time, especially when I do an all search (which by the way, I have not enjoyed that I cannot turn it off on an all search).
Hit the X on the synopsis and this will be remembered when you run future searches. In this way, you can use the feature only when you want.
0 -
Alan Palmer (Logos) said:
Generating questions is even less of an issue than summarization.
Thanks Alan, that is very useful information
Alan Palmer (Logos) said:Synopsis is uniquely expensive. As I mentioned earlier, we are considering many options to help users not hit their limit from this feature. We may change the default to only execute it on demand (it's currently defaulted on and stays on until users close it). We are constantly evaluating new AI models and adjusting our limits as third-party costs change.
I think if the gauge of how much we have used, or have left, comes through it will allow us to self manage better (or make us feel like we are) and hopefully it tokens will become so cheap it becomes irrelevant to track.
Thanks again for the information.
👁️ 👁️
0 -
MJ. Smith said:Mark said:
which by the way, I have not enjoyed that I cannot turn it off on an all search).
Actually, adding an AND or OR has the effect of turning it off
MJ Smith, if I put in the word, "Timothy" in an all search, I do not see how adding AND or OR turns off AI.
If I put in "Timothy and Paul", I see AI results also.
0 -
Alan Palmer (Logos) said:
So far, the most common reason has been extensive use of resource translation. It is an extremely popular feature among a very small cohort of users. This has been true since launching the feature in L10.
Thanks, Alan, but this also confuses me. Logos 10 full features contains translation, correct? If one does not subscribe and has Logos 10 full features, wouldn't they be able to use it to their heart's content? I do work in other countries and thus the translation feature I expect to use often. Am I in danger of using up credits? And while I am a subscriber, if I were not, would it not be true that with my L10 full features I would be able to use translation?
0 -
Alan Palmer (Logos) said:
Hit the X on the synopsis and this will be remembered when you run future searches. In this way, you can use the feature only when you want.
Alan, this is helpful. Perhaps it answers my question put to MJ Smith. It seems the other results are not AI, but I can click the button on the top right corner of each hit to summarize if I choose. That would use AI. If I got this correct, then I now understand.
If this is now understood by me, then I would possibly suggest a more drawn out explanation for other users who may get confused by the AI symbol next to the hits. That is, in the ALL search, there are no templates. But an explanation can perhaps be put here or a link to how to use All Search without AI might help
0 -
Mark said:
MJ Smith, if I put in the word, "Timothy" in an all search, I do not see how adding AND or OR turns off AI.
If I put in "Timothy and Paul", I see AI results also.
MJ Smith, I figured it out. I am not seeing AI results. I am seeing results that I could summarize and if I summarized, it would be using AI. Appreciate as always your patience and help.
0 -
Mark said:
if I put in the word, "Timothy" in an all search, I do not see how adding AND or OR turns off AI.
If I put in "Timothy and Paul", I see AI results also.
using AND or OR switches the "all" search from a smart search to a precise search. Note: AND and OR must be in all caps for the system to recognize properly. If syntax is not entered in such a way as to switch the all search into a precise search, the results it shows will always utilize AI to generate. "Summarize" and "Synopsis" utilize AI credits in addition to smart search. However, the message is clear: we should have enough AI credits so that we don't have to worry much about how these credits are being deducted from our monthly account allotment.
0 -
Aaron Hamilton said:Mark said:
if I put in the word, "Timothy" in an all search, I do not see how adding AND or OR turns off AI.
If I put in "Timothy and Paul", I see AI results also.
using AND or OR switches the "all" search from a smart search to a precise search. Note: AND and OR must be in all caps for the system to recognize properly. If syntax is not entered in such a way as to switch the all search into a precise search, the results it shows will always utilize AI to generate. "Summarize" and "Synopsis" utilize AI credits in addition to smart search. However, the message is clear: we should have enough AI credits so that we don't have to worry much about how these credits are being deducted from our monthly account allotment.
And what if I want to search only for "Timothy"? If I write Timothy, the smart search stay on. If I write "Timothy AND" the message says a precise search will be run but no search results show...at all
0 -
Aaron Hamilton said:
However, the message is clear: we should have enough AI credits so that we don't have to worry much about how these credits are being deducted from our monthly account allotment.
I am sorry, but while the communication is clear, the message is not. Alan has mentioned "Admittedly, Smart Synopsis is quite "expensive" relative to the other features".
Alan has also mentioned, "Synopsis is uniquely expensive".
Alan has also mentioned in response to those who have used up their credit amount, "So far, the most common reason has been extensive use of resource translation. It is an extremely popular feature among a very small cohort of users. This has been true since launching the feature in L10"
I apparently have not come close to using up credits. But I do plan to use both Synopsis and translation a lot. And that is why I want to know what the limits are. I don't wish to abuse the so called generousity. I want to know the limits so I can wisely use Logos in the Subscription era.
0 -
Mark said:
And what if I want to search only for "Timothy"?
I'm not aware of a solution for that, but a workaround may be possible that would consist of some arbitrary syntax that would serve the sole function of switching the search to a precise search. I also would appreciate a toggle from smart to precise, just as I would appreciate a toggle from the entire Logos catalog to owned resources. For now, we'll have to utilize the "books" search for these options and have to forego the info cards that accommodate the Logos 10 all search. The smart all search is cool. But we did lose some functionality with the all search as well.
0 -
Mark said:
I apparently have not come close to using up credits. But I do plan to use both Synopsis and translation a lot. And that is why I want to know what the limits are. I don't wish to abuse the so called generousity. I want to know the limits so I can wisely use Logos in the Subscription era.
You are certainly not alone here. As far as I understand they will be addressing this problem soon.
0 -
Well, my questions are answered. I hope Alan and others are listening. The thread has been helpful. Thanks Aaron for your contributions and everyone else.
0 -
Aaron Hamilton said:Mark said:
And what if I want to search only for "Timothy"?
I'm not aware of a solution for that, but a workaround may be possible that would consist of some arbitrary syntax that would serve the sole function of switching the search to a precise search. I also would appreciate a toggle from smart to precise, just as I would appreciate a toggle from the entire Logos catalog to owned resources. For now, we'll have to utilize the "books" search for these options and have to forego the info cards that accommodate the Logos 10 all search. The smart all search is cool. But we did lose some functionality with the all search as well.
The easiest thing would be if Logos would just give us the toggle for the All search. The user shouldn't have to figure out how to trick the software into to doing something, they should have the control.
Another issue I have is that when it switches to Precise search, it also seems to stop searching the whole store. What if I want to use precise search to search the whole store and I still want the functionality of the All search tab in how it displays results? The Bookstore tab doesn't work that well and doesn't show you enough information. For example, I can no longer do a headword search in the Bookstore tab. (I reported this but was told it was working fine. It's not.) I also can't summarize search results the Bookstore tab.
So ideally we would have the ability to toggle both the Smart vs. Precise search and the Whole Store vs. Your Books search. Then we could do a headword search for the whole store and summarize the results to see if a book has the info we're looking for before we decide to buy it.
0 -
I'm not entirely sure of the underlying technology, but I have the feeling that it would be a great challenge to implement precise search on the entire Logos catalog. It would also be extremely taxing on the Logos servers I expect.Kiyah said:What if I want to use precise search to search the whole store and I still want the functionality of the All search tab in how it displays results?
0 -
Aaron Hamilton said:
I'm not entirely sure of the underlying technology, but I have the feeling that it would be extremely difficult to use precise search on the entire Logos catalog. It would also be extremely taxing on the Logos servers I expect.
But we used to be able to do this with the Bookstore tab. In fact, I can still do Precise searches in the Bookstore tab (just tried a few things). It just seems like the headword search is broken. Other stuff seems to be working.
Here's the post reporting that issue: BUG: Headword Search no longer works in the Bookstore tab - Logos Forums
0
